扫描输入彩色地图的去噪与自动分割  

Noise Removal and Automatic Segmentation of Scanned Color Map

作　　者：刘国瑞[1] 曾平[1] 

机构地区：[1]西安电子科技大学计算机学院,陕西西安710071

出　　处：《计算机应用研究》2004年第10期114-116,共3页Application Research of Computers

基　　金：国防科技预言基金资助项目(51416050301DZ0140);陕西省自然科学基金资助项目(2001x06)

摘　　要：针对扫描输入纸质彩色地图噪声干扰严重,自动分割效果不理想的现状,提出了一种先去噪后分割的地图分割新方法。根据噪声性质采用维纳滤波与小波阈值萎缩相结合的去噪方法,进行基于CIE L a b 色空间的颜色聚类。与传统的彩图分割相比,该方法对噪声具有更强的鲁棒性,处理速度快且分割清晰。The segmentation result of scanned color map is always unsatisfied due to it′s grievous noise interference.To solve the problem, a new two- step algorithm is presented in this paper.First,WaveShrink and Wiener smoothing is adapted as the main technique to remove noise.Then a concise color clustering algorithm based on probability analysis of the CIE- L~*a~*b~* color space is exploited.By contrast with the traditional segmentation algorithm, this one provides more robustness to noise, satisfying segmentation effect and faster processing velocity.

关 键 词：彩色地图分割 离散双正交小波变换 阈值萎缩 颜色聚类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]